JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY VOL. 66, NO.

23, 2015

ª 2015 BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION ISSN 0735-1097/$36.00

PUBLISHED BY ELSEVIER INC. http://dx.doi.org/10.1016/j.jacc.2015.10.017

REVIEW TOPIC OF THE WEEK

Hemodynamics of
Mechanical Circulatory Support
Daniel Burkhoff, MD, PHD,*y Gabriel Sayer, MD,z Darshan Doshi, MD,* Nir Uriel, MDz

JACC JOURNAL CME

This article has been selected as the month’s JACC Journal CME activity, end-systolic and end-diastolic pressure-volume relations and which of
available online at http://www.acc.org/jacc-journals-cme by selecting the their features can be used to index contractility and diastolic properties;
CME tab on the top navigation bar. 2) describe how changes in preload, afterload, ventricular contractility,
and heart rate impact the left ventricular pressure-volume loop
Accreditation and Designation Statement (specifically end-diastolic volume and pressure, stroke volume, systolic
pressure generation) and myocardial oxygen demand; and 3) describe
The American College of Cardiology Foundation (ACCF) is accredited by
anatomic and physiological differences between the different types of
the Accreditation Council for Continuing Medical Education (ACCME) to
mechanical circulatory support currently in use clinically.
provide continuing medical education for physicians.

The ACCF designates this Journal-based CME activity for a maximum of 1 CME Editor Disclosure: JACC CME Editor Ragavendra R. Baliga, MD, FACC,
AMA PRA Category 1 Credit(s). Physicians should only claim credit has reported that he has no financial relationships or interests to disclose.
commensurate with the extent of their participation in the activity.
Author Disclosures: Dr. Burkhoff is an employee of HeartWare Interna-
Method of Participation and Receipt of CME Certificate tional; a consultant to Sensible Medical (founder of PVLoops LLC), and
Corvia (hemodynamic core lab director); and has received grant support
To obtain credit for JACC CME, you must:
from Abiomed. Dr. Doshi has received grant support from Abiomed. Dr.
1. Be an ACC member or JACC subscriber.
Uriel is a consultant to Thoratec, Heartware International, and Abiomed;
2. Carefully read the CME-designated article available online and in this
and has received grant support from HeartWare International. Dr. Sayer
issue of the journal.
has reported that he has no relationships relevant to the contents of this
3. Answer the post-test questions. At least 2 out of the 3 questions
paper to disclose.
provided must be answered correctly to obtain CME credit.
4. Complete a brief evaluation.
Medium of Participation: Print (article only); online (article and quiz).
5. Claim your CME credit and receive your certificate electronically by
following the instructions given at the conclusion of the activity.
CME Term of Approval

CME Objective for This Article: After reading this article, the reader Issue Date: December 15, 2015
should be able to: 1) describe the characteristics of the left ventricular Expiration Date: December 14, 2016

Listen to this manuscript’s

audio summary by
JACC Editor-in-Chief From the *Division of Cardiology, Columbia University, New York, New York; yHeartWare International, Framingham, Massa-
Dr. Valentin Fuster. chusetts; and the zDepartment of Medicine, University of Chicago, Chicago, Illinois. Dr. Burkhoff is an employee of HeartWare
International; a consultant to Sensible Medical (founder of PVLoops LLC), and Corvia (hemodynamic core lab director); and has
received grant support from Abiomed. Dr. Doshi has received grant support from Abiomed. Dr. Uriel is a consultant to Thoratec,
Heartware International, and Abiomed; and has received grant support from HeartWare International. Dr. Sayer has reported that
he has no relationships relevant to the contents of this paper to disclose.

Manuscript received July 27, 2015; revised manuscript received September 14, 2015, accepted October 2, 2015.
2664 Burkhoff et al. JACC VOL. 66, NO. 23, 2015

Hemodynamics of Circulatory Support DECEMBER 15, 2015:2663–74

Hemodynamics of
Mechanical Circulatory Support

ABSTRACT

An increasing number of devices can provide mechanical circulatory support (MCS) to patients with acute hemodynamic
compromise and chronic end-stage heart failure. These devices work by different pumping mechanisms, have various
flow capacities, are inserted by different techniques, and have different sites from which blood is withdrawn and returned
to the body. These factors result in different primary hemodynamic effects and secondary responses of the body.
However, these are not generally taken into account when choosing a device for a particular patient or while
managing a patient undergoing MCS. In this review, we discuss fundamental principles of cardiac, vascular, and pump
mechanics and illustrate how they provide a broad foundation for understanding the complex interactions between
the heart, vasculature, and device, and how they may help guide future research to improve patient outcomes.
(J Am Coll Cardiol 2015;66:2663–74) © 2015 by the American College of Cardiology Foundation.

F or patients with advanced heart failure, there

are an increasing number of therapies, espe-
cially in the form of mechanical circulatory
support (MCS). There are several classes of MCS de-
basics of ventricular mechanics, ventricular-vascular
coupling, and myocardial energetics (see [8–10] for
detailed descriptions). We will then review how these
principles can be applied to better understand the
vices, distinguished by hemodynamic characteristics hemodynamic effects of MCS.
of the pump, the sites from which blood is withdrawn
and returned, the size of catheters and/or cannulas FUNDAMENTALS OF
used, whether the insertion technique is percuta- LEFT VENTRICULAR MECHANICS
neous or surgical, and whether or not a gas exchange
unit is used. Some devices are for short-term Events occurring during a single cardiac cycle are
use, whereas others can be used for the duration of depicted by ventricular pressure–volume loops (PVLs)
a patient’s life. These characteristics contribute to (Figure 1A). Under normal conditions, the PVL is
determining the ease of deployment, ease of patient roughly trapezoidal, with a rounded top. The 4 sides
management while on the device, and overall safety of the loop denote the 4 phases of the cardiac cycle:
profile, as reviewed recently in detail (1). To varying 1) isovolumic contraction; 2) ejection; 3) isovolumic
degrees, all available devices improve cardiac output relaxation; and 4) filling. The loop falls within the
and blood pressure (2–5), but their specific features boundaries of the end-systolic pressure–volume
result in different overall hemodynamic effects. The relationship (ESPVR) and the end-diastolic pressure–
implications of these differences are only partially volume relationship (EDPVR). The ESPVR is reason-
understood (6) and have not yet been researched in ably linear, with slope Ees (end-systolic elastance)
clinical trials. and volume–axis intercept Vo. The EDPVR is
Right heart catheterization with a pulmonary nonlinear and described by simple equations, such as:
artery catheter (PAC) is the cornerstone of a standard P ¼ b(e a [V-Vo] – 1) or P ¼ bVa . ESPVR, and EDPVR shifts
clinical hemodynamic evaluation of patients under- occur with changes in ventricular contractility and
going MCS. However, widespread routine use of PAC diastolic properties (remodeling).
has declined over the past decade and there is no The actual position and shape of the loop depend
consensus on systematic use of PAC data (7). As a on ventricular pre-load and afterload. At the organ
result, important differences in hemodynamic effects level, pre-load can be defined as either end-diastolic
of different forms of MCS may have gone unrecog- pressure (EDP) or the end-diastolic volume (EDV),
nized. A full understanding from advanced hemo- which relate to average sarcomere stretch throughout
dynamic principles of the mechanisms of such the myocardium. Afterload is determined by the he-
differences has the potential to impact clinical prac- modynamic properties of the vascular system against
tice and outcomes. which the ventricle contracts and is most generally
This review aims to provide a concise overview of characterized by its impedance spectrum (the
advanced hemodynamic principles, including the frequency-dependent ratio and phase shift between
JACC VOL. 66, NO. 23, 2015 Burkhoff et al. 2665
DECEMBER 15, 2015:2663–74 Hemodynamics of Circulatory Support

pressure and flow, as determined by Fourier anal- appropriately indexed by its dimensionless ABBREVIATIONS

ysis). Afterload is more simply indexed by total pe- stiffness constant, defined as (dP/dV)/(P/V) AND ACRONYMS

ripheral resistance (TPR), the ratio between mean (8). For the case when the EDPVR is fit to the
CVP = central venous pressure
pressure and flow. Afterload can also be depicted on equation P ¼ bVa , it can be shown that a is the
Ea = effective arterial
the pressure–volume plane by the “effective arterial stiffness constant. Because it requires
elastance
elastance” (Ea) line (Figure 1A) (11). The slope of the measuring EDP and EDV over a range of vol-
ECMO = extracorporeal
Ea line is approximately equal to TPR/T, where TPR is umes, quantification of the stiffness constant membrane oxygenation
in units of mm Hg , s/ml and T is the duration of the can be difficult in practice, especially when EDP = end-diastolic pressure
heartbeat in seconds. The Ea line starts on the volume EDP is low and the nonlinear portion is not
EDPVR = end-diastolic
axis at the EDV and intersects the ESPVR at the ven- readily apparent. pressure–volume relationship
tricular end-systolic pressure-volume point of the Another index of diastolic properties is EDV = end-diastolic volume
PVL. This allows approximation of stroke volume (SV) ventricular capacitance (Figure 2B), the vol- Ees = end-systolic elastance
(the width of the loop) and ventricular end-systolic ume at a specified filling pressure. Capaci-
ESPVR = end-systolic
pressure (Pes) (the height of the loop). Pes is closely tance indexes the degree to which the EDPVR pressure–volume relationship
related to mean arterial pressure (MAP): MAP is either dilated (as with ventricular remod- LA = left atrial/atrium
z 0.9,Pes. When TPR, heart rate, or pre-load volume eling in chronic heart failure with reduced LV = left ventricle/ventricular
changes, the Ea line rotates and/or shifts so that its ejection fraction) or smaller than appropriate
LVAD = left ventricular assist
intersection with the ESPVR occurs at a different (as occurs in hypertrophic cardiomyopathy device
point (Figure 1B). This construct can be used to un- and other forms of diastolic heart failure). We MAP = mean arterial pressure
derstand ventricular–vascular coupling, which is the and others have used V 30 , the volume at an MCS = mechanical circulatory
science of describing how SV, MAP, and other key EDP of 30 mm Hg, as the index of ventricular support

cardiovascular parameters are determined by pre- capacitance. MVO2 = myocardial oxygen

load, afterload, and contractility (Figure 1B). Specif- In addition to providing a platform for consumption

ically, SV can be estimated according to: SV z (EDV  explaining ventricular mechanics, the pres- PAC = pulmonary artery
catheter
Vo)/(1 þ Ea/Ees). Cardiac output is obtained by sure–volume diagram also provides a plat-
multiplying SV by heart rate, and ejection fraction is form for understanding the determinants of PCWP = pulmonary capillary
wedge pressure
obtained by dividing SV by EDV. Similarly, MAP can myocardial oxygen consumption (MVO 2)
Pes = ventricular end-systolic
be estimated by: MAP z 0.9,(EDV  Vo)/(1/Ees þ 1/Ea). (Figure 3A) (13). MVO 2 is linearly related to
pressure
The ESPVR shifts with changes in ventricular ventricular pressure–volume area (PVA),
PVA = pressure–volume area
contractility (Figure 1C) (8,12). Increases and de- which is the sum of the external stroke work
RA = right atrial/atrium
creases in contractility are associated with leftward (the area inside the PVL) and the potential
RPM = rotations per minute
and rightward shifts of the ESPVR, respectively, energy. Potential energy is the area bounded
RV = right ventricle/ventricular
which are generally manifested as changes in Ees. In by the ESPVR, the EDPVR, and the diastolic
reality, Vo can also shift with changes in contractility. portion of the PVL, and represents the resid- SV = stroke volume

It is therefore necessary to account for changes of ual energy stored in the myofilaments at the TPR = total peripheral
resistance
both Ees and Vo when using ESPVR to index end of systole that was not converted to
Vo = volume–axis intercept
contractility. This can be achieved through use of an external work.
index that integrates changes in both Ees and Vo,
such as V 120 , the volume at which the ESPVR reaches APPLICATION TO MCS
120 mm Hg: V 120 ¼ 120/Ees þ Vo. Higher values of
V120 are associated with decreased contractility and Current modes of left ventricular (LV) MCS can be
vice versa. characterized by 1 of 3 different circuit configurations
The EDPVR is nonlinear and defines the passive (Central Illustration): 1) pumping from the right atrium
diastolic properties of the ventricle (Figure 2A). This (RA) or central vein to a systemic artery; 2) pumping
nonlinearity introduces complexity when indexing from the left atrium (LA) to a systemic artery; or 3)
diastolic properties, specifically diastolic stiffness. pumping from the LV to a systemic artery (generally
Stiffness is the change in pressure for a given change the aorta). Peak flow rates achievable by different
in volume (dP/dV). Accordingly, diastolic stiffness systems range from approximately 2.5 to 7.0 l/min.
varies with filling pressure, increasing as EDP Flow rates and circuit configurations both have a
increases, even in normal hearts. Some reports major impact on their overall cardiac and systemic
incorrectly quantify stiffness by the ratio of EDP to effects. Many other factors also affect the response to
EDV (P/V), which also varies with filling pressure MCS, including: 1) the cardiovascular substrate (i.e.,
(Figure 2A). From an engineering perspective, dia- whether the patient has a prior history of chronic
stolic material properties of the heart can be more heart failure with a dilated, remodeled LV and/or
2666 Burkhoff et al. JACC VOL. 66, NO. 23, 2015

Hemodynamics of Circulatory Support DECEMBER 15, 2015:2663–74

F I G U R E 1 Overview of PVLs and Relations

A 150 ESPVR
B 150 ESPVR C 150
LV Pressure (mm Hg)

LV Pressure (mm Hg)

LV Pressure (mm Hg)

Ea V120
Ea

Ba
100 100 100

TP

e
se

lin
Ro

Ees
lin

se
s

rH

Ba
e
TP Ee

R
R
50 50 or 50

Pr
HR

el
Ees EDPVR

oa
d
0 0 0
0 Vo 50 ESV 100 150 EDV 0 Vo 50 100 150 EDV 0 50 100 150
LV Volume (ml) LV Volume (ml) LV Volume (ml)

(A) Normal pressure–volume loop (PVL), is bounded by the end-systolic pressure–volume relationship (ESPVR) and end-diastolic pressure–volume rela-
tionship (EDPVR). ESPVR is approximately linear with slope end-systolic elastance (Ees) and volume–axis intercept (Vo). Effective arterial elastance (Ea) is
the slope of the line extending from the end-diastolic volume (EDV) point on the volume axis through the end-systolic pressure–volume point of the loop.
(B) Slope of the Ea line depends on total peripheral resistance (TPR) and heart rate (HR), and its position depends on EDV. (C) The ESPVR shifts with changes
in ventricular contractility, which can be a combination of changes in Ees and Vo. Changes in contractility can be indexed by V120, the volume at which the
ESPVR intersects 120 mm Hg. ESV ¼ end-systolic volume; LV ¼ left ventricular.

right ventricle [RV], or whether it is a first event, improved ventricular and vascular function. Finally,
with previously normal heart structure); 2) the de- the characteristics of the pump (e.g., pulsatile,
gree of acute LV recovery following initiation of axial, or centrifugal flow) can also have an impact
MCS (e.g., potentially recoverable in some forms of on several aspects of the hemodynamic responses
acute coronary syndrome, but less likely recover- to MCS (14).
able with idiopathic cardiomyopathy); 3) right-sided It is therefore important to understand and
factors, such as RV systolic and diastolic function distinguish between the primary hemodynamic
and pulmonary vascular resistance; 4) the degree to effects of a device (i.e., the expected effects on
which baroreflexes are intact and can modulate pressures and flow in the absence of any change in
vascular and ventricular properties; 5) concomitant native heart or vascular properties) and the net
medications; and 6) metabolic factors, such as hemodynamic effects observed after accounting for
pH and pO 2, which, if corrected, could result in the impact of secondary modulating factors invoked

F I G U R E 2 Characteristics of the EDPVR

A 40 B 40
LV Pressure (mm Hg)
LV Pressure (mm Hg)

30 30 V30

20 20
dP/dV
P/V

10 10
P/V
dP/dV
0 0
0 50 100 150 200 250 0 50 100 150 200 250
LV Volume (ml) LV Volume (ml)

(A) The EDPVR is nonlinear. Stiffness is indexed by the change in pressure divided by the change in volume (dP/dV), varies with pressure. P/V,
the ratio of end-diastolic pressure to volume, also varies with pressure. The myocardial stiffness constant, (dP/dV)/(P/V), is considered a valid
measure of myocardial diastolic material properties. (B) One clinically useful index of diastolic properties is ventricular capacitance, which is the
volume at a specified pressure such as V30, the volume at 30 mm Hg. Abbreviations as in Figure 1.
JACC VOL. 66, NO. 23, 2015 Burkhoff et al. 2667
DECEMBER 15, 2015:2663–74 Hemodynamics of Circulatory Support

F I G U R E 3 Myocardial Energetics Assessed on the Pressure–Volume Diagram

A 150 B Oxygen for:

PVA=SW+PE
0.20 Mechanical Work
Calcium Cycling
LV Pressure (mm Hg)

MVO2 (mlO2/beat)
100 Basal Metabolism
0.15

SW 0.10
50

PE
0.05
0
0 50 100 150 0 5000 10000 15000
LV Volume (ml) PVA (mm Hg.ml)

(A) Pressure–volume area (PVA) is the sum of the stroke work (SW) and potential energy (PE). (B) Myocardial oxygen consumption (MVO2) is
linearly correlated with PVA and is divided into 3 major components, as indicated in the figure. LV ¼ left ventricular.

following initiation of MCS. Both components of hemodynamic effect is increased LV afterload pres-
device effects will be discussed later. sure and effective Ea. If TPR and LV contractility are
Finally, use of the theories of ventricular me- fixed, the only way for the LV to overcome the
chanics detailed earlier within the context of a increased afterload is via the Starling mechanism, and
comprehensive cardiovascular simulation (9,10) blood accumulates in the LV. Consequently, LV EDP,
facilitates illustration and comparison of the hemo- LA pressure, and pulmonary capillary wedge pressure
dynamic effects of different forms of MCS. The (PCWP) increase, and the PVL becomes increasingly
simulation we used has been detailed, can be used to narrow (decreased native LV SV) and taller (increased
understand the physiology of MCS, and has been afterload pressure), and shifts rightward and upward
validated to a certain degree pre-clinically (15). Other along the EDPVR. Because the EDPVR is nonlinear,
aspects of validation and limitations of the simulation large increases in LV EDP may cause only subtle
have also been detailed previously (15–17). Note that increases in LV EDV. An echocardiogram showing a
the response of a given patient to MCS must account persistently closed aortic valve during ECMO would
for baseline pre-load, afterload LV contractility, and also signify a state of maximal LV loading and high
the flow rate of the MCS pump. For simplicity, sub- PCWP. These increases in LV pre-load and PCWP are
sequent comparisons keep these factors constant. detrimental to blood oxygen saturation coming from
Importantly, the basic principles to be discussed the lung and markedly increase myocardial oxygen
apply across a wide range of conditions. demand (increased PVA), which can worsen LV
RA-TO-ARTERIAL CIRCULATORY SUPPORT. Extra- function, especially in the setting of acute myocardial
corporeal venoarterial membrane oxygenation ischemia or infarction.
(ECMO), also referred to as extracorporeal life sup- These responses to ECMO can be modulated by
port, utilizes a pump with the capacity to assume secondary regulatory factors that influence either
responsibility for the entire cardiac output and a gas TPR or LV contractility. TPR can be reduced naturally
exchange unit for normalizing pCO 2, pO 2, and pH. by the baroreceptors, pharmacologically (e.g., nitro-
However, strictly on a hemodynamic basis, the use of prusside), or mechanically (e.g., by intra-aortic
this circuit configuration can cause significant in- balloon pumping). As illustrated in Figure 4B, a 50%
creases in LV pre-load and, in some cases, pulmonary reduction in TPR during ECMO markedly blunts the
edema. This is illustrated in Figure 4A, which depicts rise in LV EDP.
PVLs in a case of cardiogenic shock due to profound, Short-term improvements in LV function can also
irreversible LV dysfunction. Baseline cardiogenic modulate the rise in PCWP. LV function can be
shock conditions (PVL in black) have a high LV EDP, improved during ECMO due to increased central aortic
low pressure generation, low SV, and low ejection pressure, the improved coronary perfusion, normali-
fraction. As ECMO flow is initiated and increased zation of blood oxygen content (improved oxygen
stepwise from 1.5 to 3.0 to 4.5 l/min, the primary delivery to the myocardium), and normalization of
2668 Burkhoff et al. JACC VOL. 66, NO. 23, 2015

Hemodynamics of Circulatory Support DECEMBER 15, 2015:2663–74

C EN T RA L IL LUSTR AT I ON Mechanical Circulatory Support: 4 Options to Pump Blood Within the Cardiovascular System

Burkhoff, D. et al. J Am Coll Cardiol. 2015; 66(23):2663–74.

Although all forms of mechanical circulatory support return blood to the arterial system, they differ with respect to the site from which they draw blood. These dif-
ferences underlie differences in their hemodynamic effects. Percutaneous (A) and durable ventricular devices (B) that take blood from the LV have similar physiology.
Extracorporeal membrane oxygenation (ECMO) withdraws blood from the right atrium or venous system and utilizes a blood gas exchange unit (C). Percutaneous
devices can also achieve LA sourcing of blood (without need for a gas exchange unit) (D). LA ¼ left atrium/atrial; LV ¼ left ventricle/ventricular.

acid-base and other metabolic abnormalities. Phar- due to their independent effects to increase MVO 2 and
macological enhancement of contractility (e.g., by potential effects on heart rate and arrhythmias. As
b-agonists or phosphodiesterase inhibitors) is also illustrated in Figure 4B, a 50% increase in LV Ees during
possible, but may not be beneficial in cardiogenic shock ECMO also blunts the primary rise in LV EDP.
JACC VOL. 66, NO. 23, 2015 Burkhoff et al. 2669
DECEMBER 15, 2015:2663–74 Hemodynamics of Circulatory Support

include atrial septostomy (to allow left-to-right

F I G U R E 4 Ventricular Effects of ECMO
shunting), a surgically placed LV vent, an intraaortic
balloon pump, or use of a percutaneous LV-to-aorta
A 125 Baseline CGS
ventricular-assist device (described in later text) (1,18).
ECMO 1.5 L/min
ECMO 3.0 L/min Incorporation of a gas exchange unit that normal-
100 ECMO 4.5 L/min
izes blood gases is a key feature of ECMO, compared
Pressure (mm Hg)

with other forms of MCS. It is important to note that

75
blood gases measured near the site of blood return do
50 not necessarily reflect blood gases throughout the
body. If, for example, blood is returned to the femoral
25 or iliac artery and pulmonary edema compromises
native lung function, oxygen delivery to the lower
0 extremities may be normal, although oxygen delivery
120 140 160 180 200
to the head and upper extremities may be signifi-
Volume (ml)
cantly compromised.
In summary, hemodynamic responses to ECMO are
B 125 Baseline CGS
complex and variable among patients due to a host of
RA–Ao 4.5 L/min
TPR 75%
100
clinical factors. In some patients, it becomes readily
TPR 50%
apparent that afterload reduction or mechanical LV
Pressure (mm Hg)

75 unloading is required, either when pulmonary edema

appears on a chest x-ray or PCWP is noted to be
50 elevated. Variable secondary effects of ECMO on TPR
and LV contractility can explain the variability of
25 responses among patients. However, even in the
presence of relatively large secondary effects, ECMO
0 by itself may not lead to significant LV unloading.
120 140 160 180 200
Volume (ml)
LA-TO-ARTERIAL CIRCULATORY SUPPORT. Tempo-
rary LA-to-arterial MCS can be achieved with
C 150 Baseline CGS
extracorporeal devices, such as TandemHeart (Car-
ECMO 4.5 L/min
Ees 125% diacAssist, Pittsburgh, Pennsylvania), which has a
125
Ees 150%
flow capacity up to w5 l/min. LA-to-arterial MCS has
Pressure (mm Hg)

100
also been investigated for long-term use in patients
75 with severe, but stable (INTERMACS $4) chronic
heart failure (19). The site of blood return is typically 1
50 or both femoral arteries for the percutaneous
approach, and the right subclavian or axillary artery
25
for the chronic application. Given that blood is
0 withdrawn directly from the LA, PCWP and LV EDP
120 140 160 180 200
decrease with this approach. In the case that the
Volume (ml)
patient has pulmonary edema, blood oxygenation can
be improved due to the reduction in PCWP. As for
(A) Impact of extracorporeal membrane oxygenation (ECMO) on
ECMO, blood must exit the LV through the aortic
pressure–volume loops, showing flow-dependent increases of
end-diastolic pressures (EDPs), increases of effective arterial
valve with LA-to-arterial MCS. Therefore, if arterial
elastance, and decreases in LV stroke volume. ECMO-dependent pressure is increased during MCS, LV pressure gen-
increases in EDP can be partially mitigated by decreases in TPR eration must also increase. In contrast to ECMO, the
(B), and/or improvements in Ees (C). CGS ¼ cardiogenic shock; necessary increase in LV pressure generation can be
RA-Ao ¼ right atrium to aorta; other abbreviations as in Figure 1.
achieved by an isolated increase in end-systolic
volume (Figure 5A). Thus, PVA and MVO 2 can be
unchanged or decreased by this approach.
When secondary factors are insufficient to self- These primary effects are modified when secondary
mitigate a rise in LV EDP, other strategies can be factors result in decreases in TPR and increases of
utilized to reduce possible increases in afterload Ees. In such cases, end-systolic and end-diastolic
pressure and allow for LV decompression. These volumes can both decrease, along with PVA and
2670 Burkhoff et al. JACC VOL. 66, NO. 23, 2015

Hemodynamics of Circulatory Support DECEMBER 15, 2015:2663–74

devices can, in principle, reach mean flows of

F I G U R E 5 Ventricular Effects of LA-to-Arterial MCS
w5 l/min. Durable devices include the HeartMate II
(St. Jude Medical, St. Paul, Minnesota) and the HVAD
A 125
Baseline CGS
LA–Ao 1.5 L/min (HeartWare, Framingham Massachusetts), and a
LA–Ao 3.0 L/min number of other devices currently under clinical
100 LA–Ao 4.5 L/min
evaluation (e.g., HeartMate III and MVAD). These
Pressure (mm Hg)

devices can reach mean flows over 7 l/min. Although

75
these devices employ different mechanisms to pump
blood (e.g., axial, centrifugal, and mixed-flow pump
50
technologies), are implanted with different tech-
niques, and have different flow capacities, the same
25
principles govern their hemodynamic effects.
Continuous pumping of blood directly from the LV,
0
120 140 160 180 200 independent of the phase of the cardiac cycle, results
Volume (ml) in loss of the normal isovolumic periods. This trans-
forms the PVL from its normal trapezoidal shape to a
B 125
Baseline CGS
LA–Ao 4.5 L/min triangular shape (Figure 6). Unlike the other forms of
+Ees 125%, +TPR 75% support, removal of blood from the LV is not depen-
100 dent on ejection through the aortic valve. As pump
Pressure (mm Hg)

flow rate increases, the LV becomes increasingly

75 unloaded (progressive leftward shifted PVL), peak LV
pressure generation decreases, and there are marked
50
decreases in PVA and MVO 2. At the same time, arterial
pressure increases, such that peak LV pressure and
25
arterial pressure are increasingly dissociated
(Figure 6B to 6E). This direct unloading also results in
0
120 140 160 180 200 decreased LA and wedge pressures. As illustrated in
Volume (ml) the cases described earlier, improved blood oxygen-
ation, systemic pressures, and perfusion may
(A) Flow-dependent changes in pressure–volume loops with left improve the metabolic milieu and invoke beneficial
atrial-to-aortic (LA-Ao) pumping, showing reducing end-diastolic secondary changes in LV contractility and TPR. For
pressures, increasing end-systolic volume, and decreasing LV the case of LV-to-arterial pumping, these secondary
stroke volume. (B) Effects on end-diastolic and end-systolic
changes result in even greater degrees of LV unload-
volumes and pressure are modified by changes in TPR and Ees.
ing (Figure 7). Also note that for this particular case of
MCS ¼ mechanical circulatory support; other abbreviations as in
Figures 1 and 4. increased LV Ees and decreased TPR, LV pressure is
sufficient to overcome aortic pressure, and LV ejec-
tion occurs; nevertheless, the more triangular shape
of the PVL is still present.
MVO 2 (Figure 5B). Because these responses can vary Another consideration for durable LV-to-arterial
significantly between patients, the net impact of LA- MCS is the difference in characteristics between
to-arterial MCS, like ECMO, can vary between patients. axial and centrifugal flow pumps, typified by the
LV-TO-AORTA CIRCULATORY SUPPORT. Several HVAD and HeartMate II, respectively, the 2 pumps in
devices pump blood from the LV to the arterial sys- most common use today. Some authors argue that the
tem, including percutaneous catheter-based trans- differences are significant, largely on the basis of
valvular devices for temporary use and fully theoretical considerations (14). However, in a recent
implantable, durable, LV assist devices (LVADs), study in experimental heart failure in which these
intended for long-term or permanent support. types of pumps were compared (20), the authors
Percutaneous transvalvular devices include the concluded that there were no pronounced acute
commercially available Impella 2.5, Impella CP, differences. This is consistent with our own recent
Impella 5.0, and Impella LD family of devices clinical data showing no significant differences in
(Abiomed, Danvers, Massachusetts) and the Percuta- overall hemodynamic effects of these 2 pumps (7).
neous Heart Pump (PHP, Thoratec, Pleasanton, Further work on this topic is needed because new
California, which has received CE Mark and is under pumps of both types are currently being introduced
clinical investigation in the United States). These into the clinic.
JACC VOL. 66, NO. 23, 2015 Burkhoff et al. 2671
DECEMBER 15, 2015:2663–74 Hemodynamics of Circulatory Support

F I G U R E 6 Ventricular Effects of LV-to-Arterial MCS

B 140

mm Hg
70
A 120
Baseline CGS
0 0.5s
LVAD 4.5 L/min
LVAD 6.0 L/min
C 140

90

mm Hg
LVAD 7.5 L/min
Pressure (mm Hg)

70

60
0
D 140

30

mm Hg
70

0 0
0 50 100 150 200
Volume (ml)
E 140

mm Hg
70

(A) Flow-dependent changes of the pressure-volume loop with LV-to-aortic pumping. The loop becomes triangular and shifts progressively leftward (indicating
increasing degrees of LV unloading). Corresponding LV and aortic pressure waveforms at baseline (B), 4.5 l/min (C), 6.0 l/min (D) and 7.5 l/min (E). With increased flow,
there are greater degrees of LV unloading and uncoupling between aortic and peak LV pressure generation. LVAD ¼ left ventricular assist device; other abbreviations as
in Figures 1, 4, and 5.

RIGHT HEART CATHETERIZATION As a first step towards that end, the theories and
simulations described earlier led us to propose a
Descriptions in the preceding sections focused on means of evaluating the adequacy of MCS and medi-
theoretical characterizations of primary and sec- cal therapy by simultaneous evaluation of central
ondary effects of different forms of acute and
chronic MCS through the window of the pressure– F I G U R E 7 Secondary Increases in Ees and Decreases in TPR
volume diagram. Because measurements of contin- Enhance Unloading Effects of LV-to-Aortic Pumping
uous volume signals are mainly restricted to the
clinical research setting, direct application in everyday 120
Baseline CGS
clinical practice is not feasible. Nevertheless, these LVAD 4.5 L/min
theories help to inform which data to collect and Ees 125%, TPR 75%
90
how to interpret it, not only on a general population
Pressure (mm Hg)

basis, but also potentially on a patient-by-patient

basis (17). 60
In this regard, information from standard PAC is
central for evaluating patients potentially in need of
30
MCS, for the definitive assessment of patient volume
status, adequacy of ventricular support, and for
diagnosis of potential MCS complications, including 0
pump thrombosis. A sound understanding of the 0 50 100 150 200
underlying theories reviewed earlier have helped Volume (ml)
guide our own development of patient evaluation and
management strategies that aim to make maximal use Abbreviations as in Figures 1 and 4.
of PAC-derived measures.
2672 Burkhoff et al. JACC VOL. 66, NO. 23, 2015

Hemodynamics of Circulatory Support DECEMBER 15, 2015:2663–74

venous pressure (CVP), PCWP, and cardiac index over individual data points can be coded, depending on
a range of device speeds (7). To achieve this, patients the adequacy of cardiac index (e.g., cardiac index
undergo a standardized speed ramp test in which >2.0 l/min/m2 ). Figure 8A shows examples of original
device rotations per minute (RPM) are initially tracings of RA, PA, and PCWP tracings obtained at the
decreased to the lowest recommended value and are highest and lowest speeds of a typical durable-LVAD
then increased stepwise by a standardized amount. At patient (7). As shown, the increase in RPM is associ-
each RPM, hemodynamic parameters are recorded ated with significant decreases in PA pressures and
after steady-state conditions are re-established PCWP; RA pressure is influenced significantly less.
(generally 2 to 5 min). Maximal RPM for the test is Note normal respiratory variations; it is important for
determined either by the maximal recommended proper results that readings be made at end-expiration
speed for the device, or the occurrence of hyperten- which, during spontaneous respiration, is during the
sion, suction events, or arrhythmias. CVP and PCWP phase at which pressures are highest (note that auto-
are plotted as a function of each other, and the mated computer analyses of these tracings generally

F I G U R E 8 Impact of Rotational Speed Variations of Durable LVADs on Standard PAC-Derived Hemodynamics

A 50
Lowest RPMs Highest RPMs
RA and PCWP Pressure
(mm Hg)

25

0
50
RA and PA Pressures
(mm Hg)

25

CI >= 2.2
B 30 LHF BiVF/Fluid Overload
C 30 LHF BiVF/Fluid Overload

RPM Step 0
1
PCWP (mm Hg)

20
PCWP (mm Hg)

20
2
3

4
10 10
5

Hypo 6 Hypo
RPM Step 7
NORMAL RHF NORMAL RHF
0 0
0 5 10 15 20 0 5 10 15 20
CVP (mm Hg) CVP (mm Hg)

(A) Original tracings of right atrial (RA), pulmonary artery (PA), and pulmonary capillary wedge pressures (PCWP) at lowest and highest speeds
measured during a ramp test. Note normal respiratory variations. (B) Simultaneous changes in central venous pressure (CVP) and PCWP as
speed of a ventricular assist device is increased. Five zones of this domain are detailed in the text. Symbols further code for whether cardiac
index (CI) is $2.0 l/min/m2. (C) Data from 4 different patients showing variability of responses to speed changes. B and C were modified from
Uriel et al. (7). BiVF ¼ biventricular failure; LHF ¼ left heart failure; LVAD ¼ left ventricular assist device; PAC ¼ pulmonary artery catheter;
RHF ¼ right heart failure; RPM ¼ rotations per minute.
JACC VOL. 66, NO. 23, 2015 Burkhoff et al. 2673
DECEMBER 15, 2015:2663–74 Hemodynamics of Circulatory Support

do not account for the phase of respiration and can that capitalize on standard hemodynamic measures
provide misleading results). As illustrated in founded on advanced hemodynamic theories have
Figure 8B, the CVP-PCWP diagram can be divided into the potential to help in the management of MCS
5 zones on the basis of proposed (though arbitrary) patients. Whether this approach results in improved
clinically acceptable ranges of CVP (3 to 12 mm Hg) and outcomes, compared with current guidelines for
PCWP (8 to 18 mm Hg): 1) normal; 2) right heart failure; patient management by the International Society for
3) left heart failure; 4) biventricular failure and/or fluid Heart and Lung Transplant (21), is the topic of
overload; and 5) hypovolemic zones. The test shown ongoing research. A preliminary retrospective study
consisted of 8 different RPMs (steps 0 to 7). This suggests that use of invasive hemodynamic-guided
particular patient starts with high values of CVP and optimization of RPMs and medical therapy has the
PCWP. As RPMs are increased, the CVP-PCWP point potential to improve clinical outcomes (22). As pre-
moves into the normal range, including achievement viously demonstrated, more direct application of
of an adequate cardiac index. Although pump speed hemodynamics can assist directly in device selection
adjustments on the basis of ramp test results have not and patient management.
yet been correlated with improved clinical outcomes,
SUMMARY
it is suggested that the optimal speed can be deter-
mined by identifying the speed that provides normal There is an increasing number of MCS options for
values for CVP, PCWP, and cardiac index. In this treating patients with acute and chronic hemody-
example, the speed at steps 4 and 5 would satisfy this namic compromise. The characteristics of these
condition. devices vary significantly, and underlie significant
An individual patient’s response depends on many differences in their primary hemodynamic effects and
factors, such as volume status, intrinsic RV contrac- secondary responses. Clinical data to guide optimal
tility, systemic and pulmonary vascular properties, device selection and use are currently lacking. Novel
and any coexisting valvular lesions. Thus, not every approaches utilizing standard hemodynamic mea-
patient can be brought into the normal ranges for all sures have the potential to be impactful. However,
measured values. Such deviations suggest the need for the more fundamental principles of cardiac me-
additional evaluations for definitive diagnosis and chanics, ventricular–vascular coupling, and ventric-
medical therapies. CVP-PCWP relations measured ular–vascular-device coupling reviewed herein
during ramp tests from 4 clinically stable, seemingly provide an even broader foundation for clarifying the
well-compensated patients, 47 to 74 years of age who issues and generating testable hypotheses to improve
were supported with a durable LVAD are shown in clinical outcomes. Application of these principles is in
Figure 8C (with cardiac index coded by symbol). These its infancy, but already yielding encouraging results
patients had reasonably controlled blood pressures (7). Basic principles that we identified for each mode
(70 to 95 mm Hg, as assessed by Doppler opening of MCS have been illustrated using a cardiovascular
pressure) and devices showed no evidence device simulation with a set of parameters that is represen-
thrombosis or malfunction (e.g., lactate dehydroge- tative of patients undergoing MCS. However, patients
nase values 190 to 385 U/l). One patient (red) starts in present with a vast range of combinations of cardiac,
the “left heart failure” zone at low speed and moves vascular, and metabolic characteristics; each patient
to the normal zone with increased speed. Another may be considered unique. Understanding the fun-
patient (blue) remains with low CVP and PCWP ranges damentals of ventricular-vascular-device interactions
independent of speed, suggesting a hypovolemic as summarized herein and elsewhere (8,15) provides a
state that might benefit from volume administration foundation for understanding individual patient re-
and/or reduction of diuretic therapy. A third patient sponses. In this regard, it is noteworthy that there is
(cyan) remains with elevated CVP and PCWP despite even less understanding of the physiology of MCS
increases in speed, always with adequate cardiac solutions for profound biventricular failure, including
index, suggesting a fluid overload state that would, total artificial hearts, biventricular percutaneous de-
perhaps, benefit from more diuresis. A fourth patient vices, or biventricular durable devices. The concepts
(green) remains with elevated CVP with minor effects reviewed also provide the foundation for addressing
on PCWP, suggestive of right-sided dysfunction. those complex settings.
Although applied here to patients with durable
devices, the same principles should apply to patients REPRINT REQUESTS AND CORRESPONDENCE: Dr.
receiving short-term percutaneous MCS. Daniel Burkhoff, Division of Cardiology, Columbia
The approach outlined in the preceding text illus- University, 177 Ft. Washington Avenue, New York,
trates that development of innovative approaches New York 10032. E-mail: db59@cumc.columbia.edu.
2674 Burkhoff et al. JACC VOL. 66, NO. 23, 2015

Hemodynamics of Circulatory Support DECEMBER 15, 2015:2663–74

REFERENCES

1. Rihal CS, Naidu SS, Givertz MM, et al. 2015 SCAI/ pressure-volume analysis: a guide for clinical, 17. Doshi D, Burkhoff D. Insight into
ACC/HFSA/STS clinical expert consensus state- translational, and basic researchers. Am J Physiol hemodynamic theories of heart failure patho-
ment on the use of percutaneous mechanical cir- Heart Circ Physiol 2005;289:H501–12. physiology and therapeutics from cardiovascular
culatory support devices in cardiovascular care. simulation: implications for clinical research and
9. Burkhoff D. HARVI: Cardiovascular Physiology &
J Am Coll Cardiol 2015;65:e7–26. practice. J Cardiac Fail 2015. In press.
Hemodynamics. Part I. Basic Physiological Concepts
2. Kar B, Gregoric ID, Basra SS, et al. The percu- (Version 2.0.0) [Mobile application software]. 2012. 18. Koeckert MS, Jorde UP, Naka Y, et al. Impella
taneous ventricular assist device in severe re- Updated 2014. Available at: https://itunes.apple. LP 2.5 for left ventricular unloading during
fractory cardiogenic shock. J Am Coll Cardiol 2011; com/gb/app/harvi/id568196279?mt¼8. Accessed venoarterial extracorporeal membrane oxygena-
57:688–96. October 2, 2015. tion support. J Card Surg 2011;26:666–8.

3. Seyfarth M, Sibbing D, Bauer I, et al. 19. Meyns B, Klotz S, Simon A, et al. Proof of
10. Burkhoff D. HARVI: Cardiovascular Physiology &
A randomized clinical trial to evaluate the safety concept: hemodynamic response to long-term
Hemodynamics. Part II. Advanced Physiological
and efficacy of a percutaneous left ventricular partial ventricular support with the synergy
Concepts (Version 2.0.0) [Mobile application soft-
assist device versus intra-aortic balloon pumping pocket micro-pump. J Am Coll Cardiol 2009;54:
ware]. Available at: https://itunes.apple.com/gb/
for treatment of cardiogenic shock caused by 79–86.
app/harvi/id568196279?mt¼8. Accessed October
myocardial infarction. J Am Coll Cardiol 2008;52:
2, 2015. 20. Bartoli CR, Giridharan GA, Litwak KN, et al.
1584–8.
Hemodynamic responses to continuous versus
11. Sunagawa K, Maughan WL, Burkhoff D, et al.
4. Burkhoff D, Cohen H, Brunckhorst C, et al., pulsatile mechanical unloading of the failing left
Left ventricular interaction with arterial load
TandemHeart Investigators Group. A randomized ventricle. ASAIO J 2010;56:410–6.
studied in isolated canine ventricle. Am J Physiol
multicenter clinical study to evaluate the safety
1983;245:H773–80. 21. Feldman D, Pamboukian SV, Teuteberg JJ,
and efficacy of the TandemHeart percutaneous
et al. The 2013 International Society for Heart and
ventricular assist device versus conventional 12. Sagawa K. The end-systolic pressure-volume
Lung Transplantation guidelines for mechanical
therapy with intraaortic balloon pumping for relation of the ventricle: definition, modifications
circulatory support: executive summary. J Heart
treatment of cardiogenic shock. Am Heart J 2006; and clinical use. Circulation 1981;63:1223–7.
Lung Transplant 2013;32:157–87.
152:469.e1–8.
13. Suga H. Ventricular energetics. Physiol Rev 22. Anyanwu E, Patel K, Henderson C, et al.
5. Estep JD, Vivo RP, Krim SR, et al. Echocardio- 1990;70:247–77. Normalization of hemodynamics following LVAD
graphic evaluation of hemodynamics in patients with implantation may be related to improved long
14. Moazami N, Fukamachi K, Kobayashi M, et al.
systolic heart failure supported by a continuous-flow term outcome (abstr). J Card Fail 2015;21:S99.
Axial and centrifugal continuous-flow rotary pumps:
LVAD. J Am Coll Cardiol 2014;64:1231–41.
a translation from pump mechanics to clinical prac-
6. Burkhoff D, Naidu SS. The science behind tice. J Heart Lung Transplant 2013;32:1–11.
percutaneous hemodynamic support: a review and KEY WORDS cardiac index, cardiac output,
15. Morley D, Litwak K, Ferber P, et al. Hemody- central venous pressure, extracorporeal
comparison of support strategies. Catheter Car-
namic effects of partial ventricular support in circulatory membrane oxygenation,
diovasc Interv 2012;80:816–29.
chronic heart failure: results of simulation vali- pulmonary capillary wedge pressure,
7. Uriel N, Sayer GT, Addetia K, et al. Hemody- dated with in vivo data. J Thorac Cardiovasc Surg ventricular assist devices
namic ramp tests in patients with left ventricular 2007;133:21–8.
assist devices. J Am Coll Cardiol HF 2015. In
16. Kono S, Nishimura K, Nishina T, et al. Go to http://www.acc.org/jacc-
press.
Autosynchronized systolic unloading during left journals-cme to take the CME
8. Burkhoff D, Mirsky I, Suga H. Assessment of ventricular assist with a centrifugal pump. quiz for this article.
systolic and diastolic ventricular properties via J Thorac Cardiovasc Surg 2003;125:353–60.